Anthropic Releases Claude Fable 5, Its Most Powerful AI Yet, With Cyber Safeguards

On June 9, Anthropic released Claude Fable 5, the most capable model it has ever made, generally available. It also did something unusual: it shipped one model as two products, split not by capability but by a layer of safety classifiers. Fable 5 goes to the public. Its twin, Claude Mythos 5, the same underlying model with the cyber safeguards lifted, stays locked to a vetted group of cyber

The article reports that Anthropic has released Claude Fable 5, a public "Mythos-class" model that shares the same core model as Claude Mythos 5 but adds safety classifiers that trigger fallback to Claude Opus 4.8 for certain cybersecurity, biology, chemistry, and model-distillation requests.[1][2] Claude Mythos 5, with these cyber safeguards lifted, remains restricted to vetted cyber defenders and critical infrastructure partners under Project Glasswing, and Anthropic claims extensive red-teaming and low jailbreak success.[1][2][3] From a CyberSE.AI perspective, this split-model design reduces but does not eliminate the risk of powerful capabilities being misused for offensive cyber operations, and it creates a high-value target in Mythos 5 whose access controls, monitoring, and usage policies must be rigorously governed. Organizations deploying or integrating such frontier models should implement continuous AI red teaming against the safety layer, enforce strict access segmentation for higher-privilege variants, and define explicit policies for dual-use cyber capabilities exposure.